regdiscrim - An R package that is a collection of various regularization methods for discriminant analysis and supervised learning. This is includes an implementation of Regularized Discriminant Analysis from Professor Jerome H. Friedman at Stanford.

diagdiscrim - An R package that is a collection of classification models that assume conditionally independent features among all classes. These models have been shown to have excellent classification peformance with high-dimensional microarray data and can be viewed as special cases of the Naive Bayes classifier.

errorest - An R package that provides a variety of error rate estimation methods for supervised classification. To assess classification performance, I have provided several widely known estimators, including random split / Monte-Carlo cross-validation, cross-validation, bootstrap, .632, .632+, apparent, and bolstering/smoothed error rates. Furthermore, I am planning to implement other lesser known estimators. Currently, I am working to add MapReduce support for these estimators via the RHIPE package and to add easy integration with the caret package.